From: Andi Kleen <ak@muc.de>
This patch adds machine check tainting. When a handled machine check
occurs the oops gets a new 'M' flag. This is useful to ignore machines
with hardware problems in oops reports.
On i386 a thermal failure also sets this flag.
Done for x86-64 and i386 so far.
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
---
25-akpm/arch/i386/kernel/cpu/mcheck/k7.c | 1 +
25-akpm/arch/i386/kernel/cpu/mcheck/non-fatal.c | 1 +
25-akpm/arch/i386/kernel/cpu/mcheck/p4.c | 3 +++
25-akpm/arch/i386/kernel/cpu/mcheck/p5.c | 1 +
25-akpm/arch/i386/kernel/cpu/mcheck/p6.c | 1 +
25-akpm/arch/i386/kernel/cpu/mcheck/winchip.c | 1 +
25-akpm/arch/x86_64/kernel/mce.c | 2 ++
25-akpm/include/linux/kernel.h | 1 +
25-akpm/kernel/panic.c | 3 ++-
9 files changed, 13 insertions(+), 1 deletion(-)
diff -puN arch/i386/kernel/cpu/mcheck/k7.c~x86-64-i386-add-mce-tainting arch/i386/kernel/cpu/mcheck/k7.c
--- 25/arch/i386/kernel/cpu/mcheck/k7.c~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/arch/i386/kernel/cpu/mcheck/k7.c Tue Oct 5 16:10:07 2004
@@ -54,6 +54,7 @@ static asmlinkage void k7_machine_check(
wrmsr (MSR_IA32_MC0_STATUS+i*4, 0UL, 0UL);
/* Serialize */
wmb();
+ tainted |= TAINT_MACHINE_CHECK;
}
}
diff -puN arch/i386/kernel/cpu/mcheck/non-fatal.c~x86-64-i386-add-mce-tainting arch/i386/kernel/cpu/mcheck/non-fatal.c
--- 25/arch/i386/kernel/cpu/mcheck/non-fatal.c~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/arch/i386/kernel/cpu/mcheck/non-fatal.c Tue Oct 5 16:10:07 2004
@@ -48,6 +48,7 @@ static void mce_checkregs (void *info)
/* Serialize */
wmb();
+ tainted |= TAINT_MACHINE_CHECK;
}
}
}
diff -puN arch/i386/kernel/cpu/mcheck/p4.c~x86-64-i386-add-mce-tainting arch/i386/kernel/cpu/mcheck/p4.c
--- 25/arch/i386/kernel/cpu/mcheck/p4.c~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/arch/i386/kernel/cpu/mcheck/p4.c Tue Oct 5 16:10:07 2004
@@ -40,6 +40,7 @@ static void unexpected_thermal_interrupt
{
printk(KERN_ERR "CPU%d: Unexpected LVT TMR interrupt!\n",
smp_processor_id());
+ tainted |= TAINT_MACHINE_CHECK;
}
/* P4/Xeon Thermal transition interrupt handler */
@@ -60,6 +61,7 @@ static void intel_thermal_interrupt(stru
printk(KERN_EMERG "CPU%d: Temperature above threshold\n", cpu);
printk(KERN_EMERG "CPU%d: Running in modulated clock mode\n",
cpu);
+ tainted |= TAINT_MACHINE_CHECK;
} else {
printk(KERN_INFO "CPU%d: Temperature/speed normal\n", cpu);
}
@@ -222,6 +224,7 @@ static asmlinkage void intel_machine_che
wrmsr(msr, 0UL, 0UL);
/* Serialize */
wmb();
+ tainted |= TAINT_MACHINE_CHECK;
}
}
mcgstl &= ~(1<<2);
diff -puN arch/i386/kernel/cpu/mcheck/p5.c~x86-64-i386-add-mce-tainting arch/i386/kernel/cpu/mcheck/p5.c
--- 25/arch/i386/kernel/cpu/mcheck/p5.c~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/arch/i386/kernel/cpu/mcheck/p5.c Tue Oct 5 16:10:07 2004
@@ -25,6 +25,7 @@ static asmlinkage void pentium_machine_c
printk(KERN_EMERG "CPU#%d: Machine Check Exception: 0x%8X (type 0x%8X).\n", smp_processor_id(), loaddr, lotype);
if(lotype&(1<<5))
printk(KERN_EMERG "CPU#%d: Possible thermal failure (CPU on fire ?).\n", smp_processor_id());
+ tainted |= TAINT_MACHINE_CHECK;
}
/* Set up machine check reporting for processors with Intel style MCE */
diff -puN arch/i386/kernel/cpu/mcheck/p6.c~x86-64-i386-add-mce-tainting arch/i386/kernel/cpu/mcheck/p6.c
--- 25/arch/i386/kernel/cpu/mcheck/p6.c~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/arch/i386/kernel/cpu/mcheck/p6.c Tue Oct 5 16:10:07 2004
@@ -72,6 +72,7 @@ static asmlinkage void intel_machine_che
wrmsr (msr, 0UL, 0UL);
/* Serialize */
wmb();
+ tainted |= TAINT_MACHINE_CHECK;
}
}
mcgstl &= ~(1<<2);
diff -puN arch/i386/kernel/cpu/mcheck/winchip.c~x86-64-i386-add-mce-tainting arch/i386/kernel/cpu/mcheck/winchip.c
--- 25/arch/i386/kernel/cpu/mcheck/winchip.c~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/arch/i386/kernel/cpu/mcheck/winchip.c Tue Oct 5 16:10:07 2004
@@ -19,6 +19,7 @@
static asmlinkage void winchip_machine_check(struct pt_regs * regs, long error_code)
{
printk(KERN_EMERG "CPU0: Machine Check Exception.\n");
+ tainted |= TAINT_MACHINE_CHECK;
}
/* Set up machine check reporting on the Winchip C6 series */
diff -puN arch/x86_64/kernel/mce.c~x86-64-i386-add-mce-tainting arch/x86_64/kernel/mce.c
--- 25/arch/x86_64/kernel/mce.c~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/arch/x86_64/kernel/mce.c Tue Oct 5 16:10:07 2004
@@ -191,6 +191,8 @@ void do_machine_check(struct pt_regs * r
panicm = m;
panicm_found = 1;
}
+
+ tainted |= TAINT_MACHINE_CHECK;
}
/* Never do anything final in the polling timer */
diff -puN include/linux/kernel.h~x86-64-i386-add-mce-tainting include/linux/kernel.h
--- 25/include/linux/kernel.h~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/include/linux/kernel.h Tue Oct 5 16:10:07 2004
@@ -146,6 +146,7 @@ extern enum system_states {
SYSTEM_RESTART,
} system_state;
+#define TAINT_MACHINE_CHECK (1<<10)
#define TAINT_PROPRIETARY_MODULE (1<<0)
#define TAINT_FORCED_MODULE (1<<1)
#define TAINT_UNSAFE_SMP (1<<2)
diff -puN kernel/panic.c~x86-64-i386-add-mce-tainting kernel/panic.c
--- 25/kernel/panic.c~x86-64-i386-add-mce-tainting Tue Oct 5 16:10:07 2004
+++ 25-akpm/kernel/panic.c Tue Oct 5 16:10:07 2004
@@ -125,7 +125,8 @@ const char *print_tainted(void)
{
static char buf[20];
if (tainted) {
- snprintf(buf, sizeof(buf), "Tainted: %c%c%c",
+ snprintf(buf, sizeof(buf), "Tainted: %c%c%c%c",
+ tainted & TAINT_MACHINE_CHECK ? 'M' : ' ',
tainted & TAINT_PROPRIETARY_MODULE ? 'P' : 'G',
tainted & TAINT_FORCED_MODULE ? 'F' : ' ',
tainted & TAINT_UNSAFE_SMP ? 'S' : ' ');
_